Localization in Digital Twin MIMO Networks: A Case for Massive Fingerprinting

TL;DR

This work tackles the high data collection cost of outdoor fingerprinting localization by introducing digital twin RF maps generated via ray tracing in a precise 3D environment. The DT RF maps populate large fingerprint databases with synthetic RSS fingerprints across multiple beams and subbands, enabling online localization through probabilistic matching to real measurements without relying on LoS or multiple base stations. The approach is evaluated with ray-tracing and DeepMIMO-based channel models, demonstrating sub-meter localization in many NLoS scenarios and showing that increasing measurement diversity (beams, bands, time) boosts accuracy. While promising for scalable wide-area localization, the study notes the need for real-world validation and calibration against actual outdoor measurements and datasets.

Abstract

Localization in outdoor wireless systems typically requires transmitting specific reference signals to estimate distance (trilateration methods) or angle (triangulation methods). These cause overhead on communication, need a LoS link to work well, and require multiple base stations, often imposing synchronization or specific hardware requirements. Fingerprinting has none of these drawbacks, but building its database requires high human effort to collect real-world measurements. For a long time, this issue limited the size of databases and thus their performance. This work proposes significantly reducing human effort in building fingerprinting databases by populating them with \textit{digital twin RF maps}. These RF maps are built from ray-tracing simulations on a digital replica of the environment across several frequency bands and beamforming configurations. Online user fingerprints are then matched against this spatial database. The approach was evaluated with practical simulations using realistic propagation models and user measurements. Our experiments show sub-meter localization errors on a NLoS location 95\% of the time using sensible user measurement report sizes. Results highlight the promising potential of the proposed digital twin approach for ubiquitous wide-area 6G localization.

Figures (5)

• Figure 1: A real-world communication system and its digital replica ShuaifengDigitalTwin. For the common case of downlink measurements, the network sends reference signals using certain beams in specific subbands and time instants and the user reports back the received power measurements.
• Figure 2: Diagram of end-to-end fingerprinting-based localization system leveraging an offline-built digital twin. The twin is built from ray tracing simulations on all possible user locations on a precise 3D map of the deployment. The twin generates RF Maps for RSS values (or other simulated data) for different beams and subbands, thus replacing the human effort of real-world data collections. These DT RF Maps can be updated or calibrated with real-world information. The maps are used in near real-time with online real-world user measurements to extract location probabilities and make a localization estimate decision.
• Figure 3: Localization accuracy for every possible user position in a 2D grid at a height of 2 meters (6.5 feet). The simulation was performed for the proof of concept scenario with 6 buildings using the baseline reporting parameters: no. beams $|\mathcal{K}| = 1$, no. subbands $|\mathcal{B}|=1$, and no. of reports in time $|\mathcal{T}|=1$. BS uses a 64-antenna ULA with position and orientation represented. Positions 1 (50m from BS, LoS) and position 2 (80m from BS, NLoS) are used next.
• Figure 4: Impact of several reporting parameter combinations on localization error for position 1 (LoS). RSS measurements in more beams ($|\mathcal{K}|$), subbands ($|\mathcal{B}|$) and repeated measurements across time ($|\mathcal{T}|$) individually and jointly improve fingerprinting accuracy.
• Figure 5: Impact of reporting parameter combinations on localization error for position 2 (NLoS). Localization error seems particularly affected by more RSS measurements in distinct beams.